Analysis and visualization methods for detecting functional activation using laser speckle contrast imaging

Peng Hu; Bochao Niu; Hang Yang; Yang Xia; Donna Chen; Chun Meng; Ke Chen; Bharat Biswal

doi:10.1111/micc.12783

Analysis and visualization methods for detecting functional activation using laser speckle contrast imaging

Microcirculation. 2022 Oct;29(6-7):e12783. doi: 10.1111/micc.12783. Epub 2022 Sep 21.

Authors

Peng Hu¹, Bochao Niu¹, Hang Yang¹, Yang Xia^{1

2}, Donna Chen³, Chun Meng^{1

2}, Ke Chen^{1

2}, Bharat Biswal^{1

3}

Affiliations

¹ University of Electronic Science & Technology of China, The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Laboratory for Neuroinformation, School of Life Science and Technology, University of Electronic Science and Technology of China, Chengdu, China.
² University of Electronic Science & Technology of China, Sichuan Institute Brain Science & Brain Inspired Intelligence, Chengdu, China.
³ Department of Biomedical Engineering, New Jersey Institute of Technology, Newark, New Jersey, USA.

PMID: 36070200
DOI: 10.1111/micc.12783

Abstract

Background: Previous studies have used regional cerebral blood flow (CBF) hemodynamic response to measure brain activities. In this work, we use a laser speckle contrast imaging (LSCI) apparatus to sample the CBF activation in somatosensory cortex (S1BF) with repetitive whisker stimulation. Traditionally, the CBF activations were processed by depicting the change percentage above baseline; however, it is not clear how different methods influence the detection of activations.

Aims: Thus, in this work we investigate the influence of different methods to detect activations in LSCI.

Materials & methods: First, principal component analysis (PCA) was performed to denoise the CBF signal. As the signal of the first principal component (PC1) showed the highest correlation with the S1BF CBF response curve, PC1 was used in the subsequent analyses. Then, we used fast Fourier transform (FFT) to evaluate the frequency properties of the LSCI images and the activation map was generated based on the amplitude of the central frequency. Furthermore, Pearson's correlation coefficient (C-C) analysis and a general linear model (GLM) were performed to estimate the S1BF activation based on the time series of PC1.

Results: We found that GLM performed better in identifying activation than C-C. Additionally, the activation maps generated by FFT were similar to those obtained by GLM. Particularly, the superficial vein and arterial vessels separated the activation region as segmented activated areas, and the regions with unresolved vessels showed a common activation for whisker stimulation.

Discussion and conclusion: Our research analyzed the extent to which PCA can extract meaningful information from the signal and we compared the performance for detecting brain functional activation between different methods that rely on LSCI. This can be used as a reference for LSCI researchers on choosing the best method to estimate brain activation.

Keywords: brain functional activation; cerebral blood flow; correlation coefficient; fast Fourier transform; general linear model; laser speckle contrast imaging.

MeSH terms

Cerebrovascular Circulation*
Hemodynamics
Laser Speckle Contrast Imaging*
Laser-Doppler Flowmetry / methods
Regional Blood Flow
Veins